Controlled Evaluation of Document-Level Inter-Mention Links for Named-Entity Type Classification in Ukrainian Texts
Open Access
- 13 October 2026
- journal article
- research article
- Published by MDPI AG in Applied Sciences
- Vol. 16 (20) , 10121
Abstract
Document-level links between repeated mentions can provide contextual evidence unavailable to a local entity-type classifier, a potentially important capability for Ukrainian because of its limited annotated resources and rich inflectional morphology. However, their specific contribution is difficult to isolate because graph-based systems often simultaneously alter the classifier architecture. We evaluate named-entity type classification with known mention boundaries on NER-UK 2.0 under controlled conditions, using frozen XLM-RoBERTa-base representations, 10 optimization seeds, and 270 graph-head training runs. The evaluation design combines a no-message baseline with matched head parameterization, degree-preserving topology shuffling, simple averaging, oracle diagnostics, and controlled context-weakening interventions; no novel graph architecture is claimed. On intervention-free data, the exact-repeat graph changed macro-F1 by (95% CI ); thus, an overall benefit was not established, although the interval did not rule out a practically meaningful effect. A Ukrainian token-level lemmatization rule increased graph coverage from 34.41% to 42.07% but did not improve performance over exact repetition (, 95% CI ). When only the target mention’s context was artificially weakened, with no explicit degradation cue, graph refinement increased target accuracy by 0.0784 (95% CI ). The advantage reversed when the entire repeat component was weakened, while simple averaging performed at least as well as the learned graph head. Overall, the results support graph-based refinement as a conditional context-recovery mechanism rather than a universal replacement for local classification. The evidence remains controlled but not independently confirmatory and does not establish gains for end-to-end NER or robustness to naturally degraded context.
Keywords
Funding Information
- Faculty of Electrical and Computer Engineering, Cracow University of Technology (E-1/2026)
This publication has 26 references indexed in Scilit:
- A review of named entity recognition: from learning methods to modelling paradigms and tasksArtificial Intelligence Review, 2025
- Improving Named Entity Recognition for Low-Resource Languages Using Large Language Models: A Ukrainian Case Study2025
- A survey on Named Entity Recognition — datasets, tools, and methodologiesNatural Language Processing Journal, 2023
- Explicitly Capturing Relations between Entity Mentions via Graph Neural Networks for Domain-specific Named Entity Recognition2021
- Hierarchical Contextualized Representation for Named Entity RecognitionProceedings of the AAAI Conference on Artificial Intelligence, 2020
- Leveraging Multi-Token Entities in Document-Level Named Entity RecognitionProceedings of the AAAI Conference on Artificial Intelligence, 2020
- Unsupervised Cross-lingual Representation Learning at Scale2020
- Recent Named Entity Recognition and Classification techniques: A systematic reviewComputer Science Review, 2018
- Reporting Score Distributions Makes a Difference: Performance Study of LSTM-networks for Sequence Tagging2017
- A survey of named entity recognition and classification2007